Wow, we've kept pushing our Open-R1 project, an open initiative to replicate and extend the techniques behind DeepSeek-R1 And even we were mind-blown by the results we got with this latest model we're releasing: OlympicCoder It's beating Claude 3.7 on (competitive)
MACHINE LEARNING
-
Deep Learning Decodes Brain Signals to Identify ADHD
By
–
Deep Learning Decodes Brain Signals to Identify ADHD!
-
Google Releases Gemma 3 in Four Sizes
By
–
Google released Gemma 3, SOTA open model in 4 different sizes: 1B, 4B, 12B, 27B https://t.co/UFFPsYE7pg pic.twitter.com/cR0uT8NDsc
— 🚨 AI News | TestingCatalog (@testingcatalog) 12 mars 2025Google released Gemma 3, SOTA open model in 4 different sizes: 1B, 4B, 12B, 27B
-
AI Now Writes Top-Tier Papers and Self-Optimizes
By
–
AI can now write papers worthy of top-tier academic journals, test hypotheses, and optimize itself.
-
ML Engineer Recruitment for European Defense Drone Technology
By
–
If you're a machine learning engineer passionate about European defense, and if you believe (like I do) that mass-produced semi-autonomous drones are the future of warfare, send your resume to talent@harmattan.ai. Harmattan is a very unique startup that is completely vertically
-
AI-Generated Paper Passes ICLR Workshop Peer Review Successfully
By
–
This was a fun experiment we ran while developing The AI Scientist-v2. With the permission of ICLR, we submitted an AI-generated paper to an ICLR workshop that passed the peer-review process. We documented the entire process and our learnings in a blog: https://
sakana.ai/ai-scientist-f
irst-publication/
… -
Training Compute Costs and AI Emissions: Research Data Analysis
By
–
That paper is more about training compute costs and emissions than about inference. Unfortunately, lots of published work in the training emissions area is rife with misinformation. This paper I and others wrote has more data on this topic:
-
Orakl Oncology’s DINOv2 Model Implementation from Meta FAIR
By
–
We shared more on Orakl Oncology's work with the open source DINOv2 model from Meta FAIR in this post
-
AI Model Accelerates Cancer Therapy Research Through Organoid Analysis
By
–
AI is helping researchers identify therapies for cancer patients. @orakldotbio trained our DINOv2 model on organoid images to more accurately predict patient responses in clinical settings. This approach outperformed specialized models and is helping accelerate their research. pic.twitter.com/7ptP6RcyL3
— AI at Meta (@AIatMeta) 11 mars 2025AI is helping researchers identify therapies for cancer patients. @orakldotbio trained our DINOv2 model on organoid images to more accurately predict patient responses in clinical settings. This approach outperformed specialized models and is helping accelerate their research.
-
Stanford Research Challenges AI Fairness Assumptions With New Benchmarks
By
–
A new study by @StanfordHAI and RegLab scholars challenges assumptions on AI fairness: ignoring differences may actually make AI systems less fair. In @techreview
, the research team introduced two new benchmarks to address this nuance: https://
technologyreview.com/2025/03/11/111
3000/two-new-measures-show-where-ai-models-fail-on-fairness/
…